Maxwell Johnson

I’m a Staff Machine Learning and Infrastructure Engineer with 13 years of experience designing large‑scale distributed systems and high‑performance ML platforms. My work focuses on optimizing inference infrastructure, GPU utilization, and end‑to‑end reliability for production AI workloads. At Anthropic, I led efforts to reduce inference latency by 47% and triple throughput through distributed batching, CUDA optimization, and Kubernetes‑based autoscaling. Before that, at Plaid, I built real‑time ML risk‑scoring systems and data pipelines supporting millions of financial transactions daily. My background also includes developing low‑latency storage systems at AWS. I specialize in Python, C++, and Go, with deep experience in PyTorch, TensorRT, Triton, and Kubernetes. I thrive in remote, high‑impact environments where performance, scalability, and reliability matter most. I’m looking for roles that combine deep systems engineering with applied machine learning at scale.